该文提出了一种基于运动估计的自适应时空视频去噪算法。算法直接嵌入在视频压缩编码中。空域上通过局部相似性寻找候选滤波点避免由噪声干扰形成的伪相似点,提高滤波点的有效性。时域上对各像素点进行运动预测,既增加了滤波点数量,减弱帧与帧之间的闪烁现象,又避免了因帧间运动而带来的运动模糊。另一方面,通过噪声强度和图像边缘信息自适应调节滤波窗口大小及滤波权重大小,实现了滤波强度自适应和空间细节的保护。
A robust spatio-temporal video denoising algorithm is proposed based on the motion estimation, which is embedded into the video encoder. On one hand, this algorithm can be used in the space domain to eliminate the false similar pixels from neighbor pixels and improve the efficiency of pixels; it also can be used in the time domain to reduce the flicker among frames, and avoid motion blur. On the other hand, the algorithm can realize the adaptive strength of the filter and protect the special detail, when the size or the weight of the filters is adaptively adjusted by the strength of the noise and the edge information.